var fs = require('fs');
-
-export type Some<T> = { _tag: "Some"; value: T };
-export type None = {_tag: "None"};
-
+import jsTokens from "js-tokens";
+import * as util from 'util';
/**
- * wrap a x in a `Some(T)`
- * @param x : variable to be wrapped.
- * @returns wrapped `x`.
+ *
+ * # REPRESENTATION
*/
-function toSome<T>(x: T): Some<T>{
- return { _tag: "Some", value: x};
-}
/**
- * @description Like the `Some(a)` and `None` in Rust.
- *
- * @example
- * ```ts
- * let exam1 : Maybe<Number> = { _tag: "Some", value: 12 };
- * let exam2 : Maybe<Number> = None;
- * ```
+ * convert a `tkTree` AST to S-expr string
+ * @param t the `tkTree`
+ * @returns S-expr String
*/
-export type Maybe<T> = Some<T> | None;
+export function tkTreeToSExp(t: tkTree): string{
+ var str = "";
+ if (Array.isArray(t)){
+ let strArray = t.map((x)=>tkTreeToSExp(x));
+ str = "(" + strArray.join(" ") + ")";
+ }else{
+ if (t=== undefined){
+ str = "%undefined"
+ }else{
+ str = t.value;
+ }
+ }
-/**
- * @description
- * the pair of the string to be matched later and the string that have been matched
- * @param matched : string have been matched
- * @param remained : string will be tested whether it'll be matched.
- */
-export type MatcheePair = {matched : string; remained : string};
+ return str;
+}
+/**inspect the inner of the representation. */
+let repr = (x : any)=>{return util.inspect(x, {depth: null})};
/**
- * @description
- * it returns a function which test if the first char of the `remained` part of
- * the argument of the function is `c`, if it's true, update the `MatchedPair` wrapped
- * in `Some`. Otherwise, it returns `None`.
- * * @param c : the char to be test.
- * @returns the updated `MatchedPair` wrapped in `Some(x)` or `None`.
+ *
+ * # TYPES
*/
-export function match1Char(c : string) : (m: MatcheePair) => Maybe<MatcheePair> {
- return (m : MatcheePair)=>{
- if (m.remained.length == 0){
- return { _tag: "None" };
- }
- const charToBeMatched = m.remained[0];
- if (charToBeMatched === c){
- return {_tag: "Some", value :{
- matched : m.matched + charToBeMatched,
- remained : m.remained.substring(1)}};
- }
- else{
- return {_tag: "None"};
- }
- }
-};
/**
+ * TokenPair for tokens' parser combinator
+ *
+ * matched: the matched (now and before) tokens
*
- * @param m : the `MatcheePair` to be consumed.
- * @returns if the length of `m.remained` >= 1; consumes the matchee by 1 char and wraps it in `Some`,
- * otherwise, returns `None`.
+ * remained: tokens to be matched
+ *
+ * ast: abstract syntax tree
*/
-export function matchAny(m : MatcheePair) : Maybe<MatcheePair>{
- if (m.remained.length >= 1){
- return {_tag: "Some", value :{
- matched : m.matched + m.remained[0],
- remained : m.remained.substring(1)}};
- }else{
- return {_tag: "None"};
- }
+export interface TokenPair {
+ matched: jsTokens.Token[]
+ remained: jsTokens.Token[]
+ ast : tkTree[]
}
+export type Some<T> = { _tag: "Some"; value: T };
+export type None = { _tag: "None" };
+export type Maybe<T> = Some<T> | None;
+
+type Token = jsTokens.Token;
+type tkTree = Token | tkTree[];
/**
- * @description
- * it returns a function which test if the first char of the `remained` part of
- * the argument of the function is between `l` and `u`, if it's true, update the `MatchedPair` wrapped
- * in `Some`. Otherwise, it returns `None`.
- * * @param l : lower bound char, 1-char string
- * * @param u : upper bound char, 1-char string
- * @returns the updated `MatchedPair` wrapped in `Some(x)` or `None`.
+ *
+ * # PARSER UNITS
*/
-export function matchRange(l : string, u : string) : (m: MatcheePair) => Maybe<MatcheePair> {
- let lCodepoint = charToCodepoint(l);
- let uCodepoint = charToCodepoint(u);
- if (l > u){
- throw new Error("Error: the codepoint of `"+l+"` is not smaller than `"+u+"`)");
- }
- return (m : MatcheePair)=>{
- if (m.remained.length < 1){
- return {_tag : "None"};
- }
- const charToBeMatched = m.remained[0];
- const codePointToBeMatched = charToCodepoint(charToBeMatched);
- if (codePointToBeMatched >= lCodepoint && codePointToBeMatched <= uCodepoint){
- return {_tag: "Some", value :{
- matched : m.matched + charToBeMatched,
- remained : m.remained.substring(1)}};
- }
- else{
- return {_tag: "None"};
- }
- }
-};
+function toSome<T>(x:T): Maybe<T>{
+ return {_tag: "Some", value: x};
+}
/**
- * convert the one-char string to codepoint.
- * @param s : the string to code point.
- * @returns if `s.length > 1` return error; otherwise, return the codepoint of `s`.
+ * like `m ==> f` in ocaml
+ * @param m matchee wrapped
+ * @param f matching function
+ * @returns wrapped result
*/
-export function charToCodepoint(s : string): number{
- if (s.length > 1){
- throw new Error("Error: the length of input string for "+s+ "is "+s.length+`,
- however, it should be 1.`);
+function thenDo(m : Maybe<TokenPair>, f : Function){
+ if (m._tag == "None"){
+ return m;
}else{
- return s.charCodeAt(0);
+ var a : Maybe<TokenPair> = f(m.value);
+ if (a._tag == "Some"){
+ a.value.ast = m.value.ast.concat(a.value.ast);
+ }
+
+ return a;
}
}
-
/**
- * @description thendo(input, f, ...) like
- * a ==> f
- * @param input: the wrapped input.
- * @param f: the function to be applied.
*
- * @returns:the applied wrapped result `MatcheePair`.
+ * @param m : the `TokenPair` to be consumed.
+ * @returns if the length of `m.remained` >= 1; consumes the matchee by 1 token
+ * and wraps it in `Some`,
+ * otherwise, returns `None`.
*/
-export function thenDo<T>(input : Maybe<T>, f : Function) : Maybe<T>{
- if (input._tag == "None"){
- return input;
- }
- else{
- let inner = input.value;
- return f(inner);
+export function matchAny(m: TokenPair): Maybe<TokenPair> {
+ if (m.remained.length >= 1) {
+ return {
+ _tag: "Some", value: {
+ matched: m.matched.concat(m.remained[0]),
+ remained: m.remained.slice(1),
+ ast : [m.remained[0]],
+ }
+ };
+ } else {
+ return { _tag: "None" };
}
}
/**
- * @description "or", like the regex `( f1 | f2 )` .
- * It returns a function `f` of which the argument is`x`.
- * if `f1(x)` is None, then `f` returns `f2(x)`. Otherwise,
- * `F` returns `f1(x)`.
- * @param f1 : 1st function to be compared
- * @param f2 : 2nd function to be compared
- * @returns:the combined function
+ * like `f1 | f2` in regex
+ * @param f1 the first tried function
+ * @param f2 the second tried function
+ * @returns wrapped result
*/
-export function orDo<T>(f1 : Function, f2: Function) : (x : T ) => Maybe<T>{
- return (x) => {
- let f1x : Maybe<T> = (f1(x));
- {
- if (f1x._tag == "None"){
- return f2(x);
- }
- else{
- return f1x;
- }
+function orDo(f1 : Function, f2 : Function){
+ return (x : TokenPair) =>{
+ let res1 : Maybe<TokenPair> = f1(x);
+ if (res1._tag == "Some"){
+ return res1;
+ }else{
+ let res2 : Maybe<TokenPair> = f2(x);
+ return res2;
}
- };
+ }
}
-
/**
-* @description repeating matching function `f`
-* zero or more times, like the asterisk `*` in regex `f*` .
-* @param f : the function to be repeated 0+ times.
-* @returns:the combined function
-*/
-export function zeroOrMoreDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- var wrapped_old_x : Maybe<T> = {_tag: "Some", value : x};
- var wrapped_new_x : Maybe<T> = wrapped_old_x;
-
- while (wrapped_new_x._tag != "None"){
- wrapped_old_x = wrapped_new_x;
- wrapped_new_x = thenDo(wrapped_old_x, f);
- };
-
- return wrapped_old_x;
- };
-}
-
-/**
-* @description Not. like the `^` inside regex of [^f].
-* returns a function `F(x)` such that if `f(x)` is `None`,
-* returns the x consuming a char; if `f(x)` is not None, F(x)
-* returns `None`.
-* @param f: the function forbidden to be matched.
-* @returns: combined function `F`.
-*/
-export function notDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- let wrapped_x : Maybe<T> = {
- _tag : "Some",
- value : x
- };
- let f_x = thenDo(wrapped_x, f);
-
- if (f_x._tag != "None"){
+ * like regex [^c]
+ * @param f input token function. one token only.
+ * @returns combined finction
+ */
+function notDo(f : Function){
+ return (x : TokenPair) =>{
+ let res1 : Maybe<TokenPair> = f(x);
+ if (res1._tag == "Some"){
return {_tag:"None"};
}else{
- return thenDo(wrapped_x, matchAny);
+ let res2 = matchAny(x);
+ return res2;
}
- };
+ }
}
+function matchToken(typeName : string, value? : string):
+ (t : TokenPair) => Maybe<TokenPair>{
+ return (t)=>{
+ let headToken = t.remained[0];
+ if (headToken.type != typeName){
+ return {_tag:"None"};
+ }else{
+ if (value === undefined || value == headToken.value){
+ let newTokenPair = {
+ matched: t.matched.concat(headToken),
+ remained: t.remained.slice(1),
+ ast : [headToken]
+ };
+ return {_tag : "Some", value : newTokenPair};
+ }else{
+ return {_tag:"None"};
+ }
+ };
+ }
+};
+
+
/**
- * if `x` is matched by `f` once, returns `f(x)`. Otherwise,
- * returns x
- * similar to `?` in regex `f?`.
- * @param f : the function to be matched
- * @returns return wrapped f(x)
+ *
+ * # TEST
*/
-export function zeroOrOnceDo<T>(f : Function): (x : T) => Maybe<T>{
- return (x)=>{
- var wrapped_old_x : Maybe<T> = {_tag: "Some", value : x};
- var wrapped_new_x = thenDo(wrapped_old_x, f);
+const tokens = Array.from(jsTokens(
+`import foo from\t 'bar';
+import * as util from 'util';
- if (wrapped_new_x._tag != "None"){
- return wrapped_new_x;
- }else{
- return wrapped_old_x;
- }
- };
-}
+花非花,霧\\{非霧 。{{foo();}}下
+一句`));
-export function tokenize(input : string){
- var input_matchee_pair : Maybe<MatcheePair> = toSome(
- {matched:"",
- remained: input});
+console.log("RESULT="+repr(tokens));
- // integer = ([+]|[-])\d\d?
- let integer = (x : MatcheePair) =>
- { let wrapped_x = toSome(x);
- let plusMinus = orDo(match1Char('+'), match1Char('-')); // ([+]|[-])
- let d = matchRange('0','9'); // \d
- return thenDo(thenDo(thenDo(wrapped_x,
- zeroOrOnceDo(plusMinus)),d),
- zeroOrMoreDo(d));
- }
- console.log(input+", result: ");
- console.log(thenDo(input_matchee_pair, integer));
- // TODO: id, string, space, basic operator, 3 marks: @, {, }.
-}
+var mainTokenPair : TokenPair = {
+ matched : [] ,
+ remained : tokens,
+ ast : []};
+
+let a = thenDo(thenDo(toSome(mainTokenPair), matchToken('IdentifierName')),
+ notDo(matchToken('Punctuator', ';')));
-tokenize("+123");
-tokenize("123");
-tokenize("-123");
-tokenize(" 123");
-tokenize("c123");
\ No newline at end of file
+
+console.log("RESULT="+repr(a));
+if (a._tag == "Some"){
+ console.log("SEXP="+tkTreeToSExp(a.value.ast));
+}